Anthropic loses Pentagon appeal as IPO plans loom
A U.S. appeals court upheld the Pentagon's exclusion of Anthropic's Claude from certain systems and contracts. The dispute centers on Anthropic's restrictions related to autonomous weapons and mass surveillance. The ruling comes as Anthropic prepares for a possible IPO that could value it around $2 trillion, and the company may still pursue further legal options.
The conflict followed failed negotiations in February. Anthropic had let U.S. agencies use Claude for analyzing overseas intelligence, designing military systems, and some cyberattacks, while retaining two prohibitions: weapons that choose and strike targets without human control, and large-scale monitoring of Americans. The Pentagon wanted a contract allowing every lawful use.
On September 25, a divided D.C. appellate panel ruled 2-1 that the department had enough reason to treat Anthropic as a procurement danger. The court observed that Claude had refused some government requests, including ones involving secret files and health data. An earlier ruling favoring Anthropic still stands, so more appeals remain possible.
The ruling could affect military contractors that rely on Anthropic tools for Pentagon-related work, as they may need alternatives. It may also influence how AI companies weigh safety restrictions against government contracts, potentially shaping future procurement norms. For Anthropic, the outcome could add uncertainty before a possible IPO, affecting investors and employees. The public may see broader debate over autonomous weapons and surveillance safeguards, though the court’s decision itself does not settle those policy questions.